Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model
The analysis of the near-surface air temperature is vital for many applications such as urban heat islands and climate change studies. In particular, extreme weather events are typically localized and so should the corresponding adaptation measures. However, climate scientists are often confronted w...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-09-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/18/4447 |
| _version_ | 1797577233811898368 |
|---|---|
| author | Afshin Afshari Julian Vogel Ganesh Chockalingam |
| author_facet | Afshin Afshari Julian Vogel Ganesh Chockalingam |
| author_sort | Afshin Afshari |
| collection | DOAJ |
| description | The analysis of the near-surface air temperature is vital for many applications such as urban heat islands and climate change studies. In particular, extreme weather events are typically localized and so should the corresponding adaptation measures. However, climate scientists are often confronted with the difficulty of providing reliable predictions at high spatial resolutions in the order of 1 km. We propose to train a convolutional neural network model to emulate the hourly high-resolution near-surface air temperature field simulated by the Weather Research and Forecasting (WRF) software over a period of 18 months. The model is driven by current and past lags of coarse SEVIRI land surface temperature fields. This nowcasting application implements a downscaling of the spatial resolution of the input by about a factor of four, while establishing a correlation between current and past land surface temperature maps and the current near-surface air temperature field. The U-Net variant that is proposed in this study uses regularization to prevent over-fitting and implements a novel space-time approach, where multiple time steps are fed into the model through 3D convolution layers. Besides LST, the model also uses urban density as additional static input to be able to predict the temperature more accurately in urban areas and to improve the generalizability of the trained model. The performance of our U-Net model is assessed via comparison to an MLR benchmark (ridge regularization). The model is shown to be superior on all performance metrics. It achieves a mean absolute error of 1.36 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C versus 1.49 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for benchmark (a 9% relative improvement) and a root mean square error of 1.77 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C versus 1.91 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for benchmark (a 7% relative improvement). The absolute error of the model is less than 2 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for 77% of the prediction points versus 72% for the benchmark (a 7% relative improvement). The improvement over the benchmark is even more significant during extreme hot periods. We demonstrate the generalizability of the approach by testing the trained model on unseen spatial domains. |
| first_indexed | 2024-03-10T22:06:15Z |
| format | Article |
| id | doaj.art-4ee4bb89afa74ee5aff5864bf49237bb |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T22:06:15Z |
| publishDate | 2023-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-4ee4bb89afa74ee5aff5864bf49237bb2023-11-19T12:47:49ZengMDPI AGRemote Sensing2072-42922023-09-011518444710.3390/rs15184447Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning ModelAfshin Afshari0Julian Vogel1Ganesh Chockalingam2Fraunhofer Institute for Building Physics, Fraunhoferstraße 10, 83626 Valley, GermanyFraunhofer Institute for Building Physics, Fraunhoferstraße 10, 83626 Valley, GermanyFraunhofer Institute for Building Physics, Fraunhoferstraße 10, 83626 Valley, GermanyThe analysis of the near-surface air temperature is vital for many applications such as urban heat islands and climate change studies. In particular, extreme weather events are typically localized and so should the corresponding adaptation measures. However, climate scientists are often confronted with the difficulty of providing reliable predictions at high spatial resolutions in the order of 1 km. We propose to train a convolutional neural network model to emulate the hourly high-resolution near-surface air temperature field simulated by the Weather Research and Forecasting (WRF) software over a period of 18 months. The model is driven by current and past lags of coarse SEVIRI land surface temperature fields. This nowcasting application implements a downscaling of the spatial resolution of the input by about a factor of four, while establishing a correlation between current and past land surface temperature maps and the current near-surface air temperature field. The U-Net variant that is proposed in this study uses regularization to prevent over-fitting and implements a novel space-time approach, where multiple time steps are fed into the model through 3D convolution layers. Besides LST, the model also uses urban density as additional static input to be able to predict the temperature more accurately in urban areas and to improve the generalizability of the trained model. The performance of our U-Net model is assessed via comparison to an MLR benchmark (ridge regularization). The model is shown to be superior on all performance metrics. It achieves a mean absolute error of 1.36 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C versus 1.49 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for benchmark (a 9% relative improvement) and a root mean square error of 1.77 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C versus 1.91 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for benchmark (a 7% relative improvement). The absolute error of the model is less than 2 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mo>°</mo></semantics></math></inline-formula>C for 77% of the prediction points versus 72% for the benchmark (a 7% relative improvement). The improvement over the benchmark is even more significant during extreme hot periods. We demonstrate the generalizability of the approach by testing the trained model on unseen spatial domains.https://www.mdpi.com/2072-4292/15/18/4447downscalingdeep learningensemble modelingU-NetLSTWRF |
| spellingShingle | Afshin Afshari Julian Vogel Ganesh Chockalingam Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model Remote Sensing downscaling deep learning ensemble modeling U-Net LST WRF |
| title | Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model |
| title_full | Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model |
| title_fullStr | Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model |
| title_full_unstemmed | Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model |
| title_short | Statistical Downscaling of SEVIRI Land Surface Temperature to WRF Near-Surface Air Temperature Using a Deep Learning Model |
| title_sort | statistical downscaling of seviri land surface temperature to wrf near surface air temperature using a deep learning model |
| topic | downscaling deep learning ensemble modeling U-Net LST WRF |
| url | https://www.mdpi.com/2072-4292/15/18/4447 |
| work_keys_str_mv | AT afshinafshari statisticaldownscalingofsevirilandsurfacetemperaturetowrfnearsurfaceairtemperatureusingadeeplearningmodel AT julianvogel statisticaldownscalingofsevirilandsurfacetemperaturetowrfnearsurfaceairtemperatureusingadeeplearningmodel AT ganeshchockalingam statisticaldownscalingofsevirilandsurfacetemperaturetowrfnearsurfaceairtemperatureusingadeeplearningmodel |